Welcome to 2022! Today ex_cldr_messages gets a material update that fully integrates it with gettext. The heavy lifting in both packages was done by @maennchen to whom much thanks is due.
Now you can use the ICU Message format in a normal gettext workflow.
A good description of motivations and differences can be found in this presentation by Mark Davis from Google in 2012.
The ICU message format is a more flexible format which also addresses two specific issues:
Two specific shortcomings that the ICU message format addresses:
Grammatical Gender
Many languages inflect in gender specific way. One example in French might be:
# You are the only participant for a male and female
Vous êtes le seul participant
Vous êtes la seule participante
# Married for a male and a female
Marié
Mariée
In Gettext this requires individual messages and conditional code in the application in order to present the correct message to an audience. This is compounded by the fact that some languages have more than two grammatical genders (most have two and four but some are attested with up to 20).
The ICU message format provides a mechanism (the select format that helps translator and UX designers implement a single message to easily encapsulate messages conditional on grammatical gender (or any other selector).
For the example above, the message format could be:
"{grammatical_gender, select,
male {Marié}
female {Mariée}
other {perhaps}
}"
Standardised plural rules
Although Gettext supports pluralization for messages through the Gettext.Plural module in Elixir and the Gettext functions like Gettext.ngettext/4, the plural rules for a language have to be implemented for each message. Given the wide differences in how plural forms are structured in different languages this can be a material challenge. For example:
- English has two plural forms: singular and plural
- French applies the singular rule to two values (0 and 1) and a plural form to larger groupings
- Japanese does not differentiate
- Russian has 4 categories
- Arabic has 6 categories
Since CLDR has a strong set of pluralization rules defined for ~500 locales, each of which is supported by ex_cldr, the ICU message format can reuse these pluralization rules in a simple and consistent fashion using the plural format.
